Identification of Potential Repurposable Drugs in Alzheimer's Disease Exploiting a Bioinformatics Analysis

Abstract

Alzheimer's disease (AD) is a neurologic disorder causing brain atrophy and the death of brain cells. It is a progressive condition marked by cognitive and behavioral impairment that significantly interferes with daily activities. AD symptoms develop gradually over many years and eventually become more severe, and no cure has been found yet to arrest this process. The present study is directed towards suggesting putative novel solutions and paradigms for fighting AD pathogenesis by exploiting new insights from network medicine and drug repurposing strategies. To identify new drug-AD associations, we exploited SAveRUNNER, a recently developed network-based algorithm for drug repurposing, which quantifies the vicinity of disease-associated genes to drug targets in the human interactome. We complemented the analysis with an in silico validation of the candidate compounds through a gene set enrichment analysis, aiming to determine if the modulation of the gene expression induced by the predicted drugs could be counteracted by the modulation elicited by the disease. We identified some interesting compounds belonging to the beta-blocker family, originally approved for treating hypertension, such as betaxolol, bisoprolol, and metoprolol, whose connection with a lower risk to develop Alzheimer's disease has already been observed. Moreover, our algorithm predicted multi-kinase inhibitors such as regorafenib, whose beneficial effects were recently investigated for neuroinflammation and AD pathology, and mTOR inhibitors such as sirolimus, whose modulation has been associated with AD.

Keywords: dementia; drug repurposing; network theory.

Figure A1

Dendrogram and heatmap of the drug-disease network. The network adjusted similarity values are clustered according to rows (diseases) and columns (drugs) by using Ward’s minimum variance method as linkage hierarchical clustering algorithm and by using the Euclidean distance as distance metric. Heatmap color key denotes the adjusted and normalized network similarity between drug targets and disease genes in the human interactome, increasing from blue (less similar) to red (more similar).

Figure A2

Number of repurposable drugs. Heatmap showing the number of drugs predicted to be repurposable for each disease (on the diagonal) and the number of drugs shared among each pair of diseases. Heatmap color scales with the number of drugs, increasing from white to red.

Figure 1

Workflow of the study. The input data are: (i) the human interactome download from [6], (ii) a list of drug targets acquired from DrugBank [19]; (iii) a list of disease-associated genes for Alzheimer’s disease (AD) and other diseases related to AD acquired from the Phenopedia [20] and DisGeNET [21] data sources. First, an in silico drug repurposing analysis was performed by using the SAveRUNNER algorithm to obtain candidate drugs for the understudied diseases. SAveRUNNER releases the drug-disease network, where the nodes are drugs and diseases, and a link occurs between them if the drug–disease association has been found to be statically significant and the weight of their association is the network-based similarity measure. Then, the original medical indications are retrieved from Therapeutic Target Database (TTD) and are assigned to each predicted drug. Looking at the drugs predicted by SAveRUNNER for AD, an in silico validation based on the gene set enrichment analysis (GSEA) has been performed by assigning a score that reflects the possible counteraction of the predicted drugs (drug signatures acquired from Connectivity Map—CMap [22,23]) on the AD pathophenotype (disease signatures acquired from the Gene Expression Omnibus—GEO [24]).

Figure 2

Drug–disease network. This graph shows the high-confidence predicted drug–disease associations connecting 14 diseases (bigger labelled circles) with the 1468 FDA-approved drugs (smaller circles). The node sizes are scaled with the number of genes associated to each analyzed disease. The edge color denotes the adjusted similarity between the drug targets and disease genes mapped on the human interactome, increasing from blue (less similar) to yellow (more similar). The node color refers to one of the six clusters identified by SAveRUNNER according to the label reported in the legend. For the clusters’ identification on the drug–disease network, SAveRUNNER exploits a cluster detection algorithm based on greedy optimization of the network modularity [35].

Figure 3

Distribution of the original medical indications of drugs predicted to be repurposable for AD. Bar plot of the top 30 original medical indications with at least five associated drugs, retrieved from TTD [27].